Added additional log messages in enterMPIPL function

Added additional log message in entermpipl to help
debugging in multi processor configuration

Tested: verified journal logs.

Signed-off-by: Jay

Added additional log messages in enterMPIPL function

Added additional log message in entermpipl to help
debugging in multi processor configuration

Tested: verified journal logs.

Signed-off-by: Jayanth Othayoth <ojayanth@in.ibm.com>
Change-Id: I39b1ab8d512a1cfe0362ae9b57b079a57059294f

show more ...

phal: set pdbg environment variable

pdbg needs PDBG_DTB environment variable initialisation
prior to pdbg init to set correct device tree back-end.

Tested: verified normal boot without systemd glo

phal: set pdbg environment variable

pdbg needs PDBG_DTB environment variable initialisation
prior to pdbg init to set correct device tree back-end.

Tested: verified normal boot without systemd global variable set.

Signed-off-by: Jayanth Othayoth <ojayanth@in.ibm.com>
Change-Id: I7ac70cbb1966edcb85db0a862781149a0bc0f492

show more ...

pdbg v3.3 related changes

Separate sbe api into new header libpdbg_sbe.h related
changes.

Signed-off-by: Jayanth Othayoth <ojayanth@in.ibm.com>
Change-Id: I9ec660067ddae0bcc61fde016aa764976263e5a4

meson: fix static initialization order race condition

When moving over to meson, openpower-proc-control started seg faulting
whenever it was run. It was not even making it to the main() function.

U

meson: fix static initialization order race condition

When moving over to meson, openpower-proc-control started seg faulting
whenever it was run. It was not even making it to the main() function.

Upon investigation, it was found that when moving to meson, this code
started hitting an issue where the hw procedure registration code was
hitting a common static initialization race condition. Whether you hit
this issue is pretty much at the whim of the compiler and the build
system so Automake did not hit it, but meson does.

The following has a pretty good write up on the issue and how to avoid
it:
https://isocpp.org/wiki/faq/ctors#static-init-order-on-first-use-members

This commit utilizes the solution proposed in that doc

Tested:
- Verified that openpower-proc-control built via meson now works as
expected.

Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I0ea5ef68e73fe4fa08e9cf8e4e3f1dbc8476a2de

show more ...

meson: remove old automake files

Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I86db79e9e91508a8877cac1d6496e5b06adc17c2

gitignore: ensure utest.cpp not ignored

Couldn't see test/utest.cpp in my Atom IDE. It doesn't show things that
match in the .gitignore file. Update the .gitignore to be specific on
what to ignore w

gitignore: ensure utest.cpp not ignored

Couldn't see test/utest.cpp in my Atom IDE. It doesn't show things that
match in the .gitignore file. Update the .gitignore to be specific on
what to ignore with the unit tests.

Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: Ifbbabdadad9252ac1596194129a6f76a3d6a36ab

show more ...

Fix related P10 SPI mux settings

Existing code set all 32 bits to true for the CFAM 2818
register. Which is not correct because First 4 bits are
for the SPI mux related settings in cfam 2818 registe

Fix related P10 SPI mux settings

Existing code set all 32 bits to true for the CFAM 2818
register. Which is not correct because First 4 bits are
for the SPI mux related settings in cfam 2818 register.

Fixed with correct settings based register definition

Tested: Verified in hardware

Signed-off-by: Jayanth Othayoth <ojayanth@in.ibm.com>
Change-Id: I3b9281bee1980a355cc44238dccace2441d625b6

show more ...

set CLK NE termination site based on HW VPD keyword.

VINI Record "HW" keyword Byte 0's MSB bit indicates
proc or planar type. This type value is used for choosing
termination site attribute.

Reads

set CLK NE termination site based on HW VPD keyword.

VINI Record "HW" keyword Byte 0's MSB bit indicates
proc or planar type. This type value is used for choosing
termination site attribute.

Reads the HW level VPD keyword planar VINI records,
compute SYS_CLK_NE_TERMINATION_SITE and set the
present processors ATTR_SYS_CLK_NE_TERMINATION_SITE
value.

This attribute initialisation should be happened prior
to set_ref_clock istep. Added this function before
istep execution in startHost procedure.

Tested: manually verified attribute values

Signed-off-by: Jayanth Othayoth <ojayanth@in.ibm.com>
Change-Id: I56f88f92df2e5db9d204b7b3cc87f15d6759d07d

show more ...

Added openpower util file support

Created common file to keep all the common util functions.

In this commit
- added getService function to get d-bus service for given
interface.
- updated crea

Added openpower util file support

Created common file to keep all the common util functions.

In this commit
- added getService function to get d-bus service for given
interface.
- updated create_pel.cpp to start using common util interface.
- Moved formatting library to common build dependency for
both meson and automake.

Tested: Manually verified.

Signed-off-by: Jayanth Othayoth <ojayanth@in.ibm.com>
Change-Id: I39a642710c0476e81fae020eb3b4216ca8ce9ad5

show more ...

use new sdbus++ camelcase for NMI.

Change I17a8d7479556596a3cf252b3f4eae9c8df547189 will change
how sdbus++ generates names which start with an acronym.
Prepare for this by keying off the SDBUSPP_NE

use new sdbus++ camelcase for NMI.

Change I17a8d7479556596a3cf252b3f4eae9c8df547189 will change
how sdbus++ generates names which start with an acronym.
Prepare for this by keying off the SDBUSPP_NEW_CAMELCASE
define to use the new format.

Changes:
nMI() -> nmi()

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I03c7c98cb784da2343719aa7f47b256b1ebcdaff

show more ...

enable shellcheck

Fix shellcheck issues in gen_makefile and enable.

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I24e6290c21cd6afd02385609d96e93522fd53380

bootstrap: fix shellcheck warnings

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I86ef079c9632e5563a1a56d6a2e23aad59ecbc57

remove host-on file rm

The removal of this file was moved into the phosphor-state-manager
applications. This is the final commit in that series:
https://github.com/openbmc/phosphor-state-manager/com

remove host-on file rm

The removal of this file was moved into the phosphor-state-manager
applications. This is the final commit in that series:
https://github.com/openbmc/phosphor-state-manager/commit/4a4c1a69432dd88a4204e7a15a838aae75ae1d12

Clearing it in the host stop path in this service file caused
complexity with other paths like BMC resets with the host up and
warm reboots.

Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I4e156d3818b225529c6a6c172d7042d5ca431804

show more ...

cfam-reset: do not run if chassis power on

Due to a variety of reasons, the op-cfam-reset.service was put in the
obmc-chassis-poweron@.target. To handle situations where the BMC is
reset when the ho

cfam-reset: do not run if chassis power on

Due to a variety of reasons, the op-cfam-reset.service was put in the
obmc-chassis-poweron@.target. To handle situations where the BMC is
reset when the host is up, it's important this service not be executed
if the host is running. Issuing a CFAM reset while the host is running
will cause it to crash.

Change the service directive to look at the chassis power file to know
whether it should run or not.

Tested:
- Verified on a rainier system that host stayed up and running through a
BMC reset with this change.

Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I103defecf113e6cc8f7eba4dd9d7c47519621015

show more ...

fix dependency issue between mpreboot and stop instr

During some testing of the mpreboot function, I noticed this journal
log:

Feb 04 17:12:05 bmc systemd[1]: op-stop-instructions@0.service: Found

fix dependency issue between mpreboot and stop instr

During some testing of the mpreboot function, I noticed this journal
log:

Feb 04 17:12:05 bmc systemd[1]: op-stop-instructions@0.service: Found ordering cycle on op-enter-mpreboot@0.service/start
Feb 04 17:12:05 bmc systemd[1]: op-stop-instructions@0.service: Found dependency on obmc-host-stopped@0.target/start
Feb 04 17:12:05 bmc systemd[1]: op-stop-instructions@0.service: Found dependency on obmc-host-stopping@0.target/start
Feb 04 17:12:05 bmc systemd[1]: op-stop-instructions@0.service: Found dependency on op-stop-instructions@0.service/start
Feb 04 17:12:05 bmc systemd[1]: op-stop-instructions@0.service: Job obmc-host-stopped@0.target/start deleted to break ordering cycle starting with op-stop-instructions@0.service/start

The issue is that op-enter-mpreboot@.service says it has to run after
the host has been stopped, but before op-stop-instructions@.service.
This is because op-enter-mpreboot@.service creates the file which tells
op-stop-instructions@.service to not run. op-stop-instructions@.service
is what causes the host to stop, so we end up with a cyclical
dependency.

To fix this, just have op-enter-mpreboot@.service run during the same
target as op-stop-instructions@.service. The only service run here in
that window is op-stop-instructions@service anyway.

This was all a result of the request we not run
op-stop-instructions@.service in the mpreboot path.

Tested: Verified I no longer see the error messages above during a
mpreboot.

Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: Ied611bad3727a11a76aae48ff0a81c30a35952b2

show more ...

add timeout to stop-instructions service

Stopping instructions on POWER based systems is a fairly complicated
process with a lot of moving parts. If one of these moving parts gets
hung up, the entir

add timeout to stop-instructions service

Stopping instructions on POWER based systems is a fairly complicated
process with a lot of moving parts. If one of these moving parts gets
hung up, the entire power off process gets hung up. This issue has been
seen intermittently over the years on different systems so for safety,
lets put a timeout in the service.

Stopping instructions is not a requirement to power off, it's just a
nice-to-have type function. Don't let it hanging prevent the overall
powering off of the system.

Note that this commit also moves the rm of the host@%i-on file to being
first. This is to ensure it always runs, even if the pdbg command times
out and causes the service to fail.

Tested:
- Verified on good path stop instructions takes 2-4 seconds.
- Verified when I caused this service to hang, we hit timeout and
powered off system.

Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I7d249051634aeca399e2ecf6d24bca5e19637160

show more ...

Execute proc-pre-poweroff after proc stop instruction

proc-pre-poweroff service shuts down SBE. This is triggering
stop instruction service failure.

This fix is to move proc-pre-poweroff after stop

Execute proc-pre-poweroff after proc stop instruction

proc-pre-poweroff service shuts down SBE. This is triggering
stop instruction service failure.

This fix is to move proc-pre-poweroff after stop instruction.

Tested:
- Patched system with service and verified correct order:
Dec 11 08:40:59 system systemd[1]: Starting Stop instructions for host0...
Dec 11 08:41:00 system systemd[1]: Finished Stop instructions for host0.
Dec 11 08:41:00 system systemd[1]: Starting Preparation to poweroff host0...
Dec 11 08:41:00 system systemd[1]: Finished Preparation to poweroff host0.

Signed-off-by: Jayanth Othayoth <ojayanth@in.ibm.com>
Change-Id: I9a305f298bd0d02bbfd83a8b6740b8898515c54f
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>

show more ...

phal: Added proc-pre-poweroff support

Changes:
-Adding new service which will be called during
chassis poweroff.

Test:
-Did poweroff to check, if the service file
is getting picked up or not and ne

phal: Added proc-pre-poweroff support

Changes:
-Adding new service which will be called during
chassis poweroff.

Test:
-Did poweroff to check, if the service file
is getting picked up or not and new added
procedure is getting called.

Signed-off-by: Chirag Sharma <chirshar@in.ibm.com>
Signed-off-by: Ramesh Iyyar <rameshi1@in.ibm.com>
Change-Id: Ic533433c4771216e5681b61cccf154f7ed029457

show more ...

cfam-reset: run before chassis power

The IBM hardware team has requested we run the CFAM reset prior to
chassis power on. This is to help mitigate some issues seen with DIMM
failures.

Tested:
- Ver

cfam-reset: run before chassis power

The IBM hardware team has requested we run the CFAM reset prior to
chassis power on. This is to help mitigate some issues seen with DIMM
failures.

Tested:
- Verified a boot completes on a rainier

Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: Id28d29e624bde4ce7c77c1e937c2c6e83363c6b3

show more ...

phal: Corrected condition to add trace for FFDC

Change-Id: I852ecaf5d2402131278fd134b2875d94b0b10e16
Signed-off-by: Ramesh Iyyar <rameshi1@in.ibm.com>

spi-mux: do not set if chassis power is on

The op-reset-chassis-running@.service determines if chassis power is on,
and if so creates the /run/openbmc/chassis@0-on file. Need to let this
service run

spi-mux: do not set if chassis power is on

The op-reset-chassis-running@.service determines if chassis power is on,
and if so creates the /run/openbmc/chassis@0-on file. Need to let this
service run and create the file if chassis power is on.

Tested:
- Verified that when BMC is rebooted with chassis power on that when
the BMC comes back up, it does not run the set-spi-mux.service.

Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I6075008dc8d7be9d111bbbc9b48606d8bb806e63

show more ...

spi-mux: ensure mux is reconfigured on chassis off

During the boot of the system, the BIOS firmware reconfigures the SPI
mux so that it has access to the P10 hardware. When the system is
powered off

spi-mux: ensure mux is reconfigured on chassis off

During the boot of the system, the BIOS firmware reconfigures the SPI
mux so that it has access to the P10 hardware. When the system is
powered off, need to switch the mux back over to the BMC so that it once
again has access to the devices behind the mux.

Continue to run this service as a part of the initial BMC boot and also
run it as a part of the chassis power off target. Ensure it runs after
chassis power has been removed when running during power off.

Tested:
- Verified that the service is now run when chassis power is turned off

Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: Ia3f38a01540961fb46b38aa0477d07882826d0b6

show more ...

build: update clang-format

Update to the latest OpenBMC clang-format.

Change-Id: I19424fe32f32900d21272e343a5bda8675edbe60
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>

meson: add checks, defaults

Default to debugoptimized builds instead of the meson default of debug
since that is the autotools default.

Anything other than c++17 is not supported, so error out if s

meson: add checks, defaults

Default to debugoptimized builds instead of the meson default of debug
since that is the autotools default.

Anything other than c++17 is not supported, so error out if something
else is selected by the user.

Make tiny builds even smaller by removing asserts and remove asserts for
release builds.

Change-Id: I034438263b1125c476dccb377dc52142415dddd2
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>

show more ...

meson: fix systemd service bugs

The logic for ENABLE_PHAL_TRUE was reversed under meson.

prefix was omitted when computing the path for the bindir replacement
token in the system unit files.

Fixes

meson: fix systemd service bugs

The logic for ENABLE_PHAL_TRUE was reversed under meson.

prefix was omitted when computing the path for the bindir replacement
token in the system unit files.

Fixes: 919e408aaad453f94e1ff3d35440b8fc8a2f3e13
Change-Id: I096d1f69554b23c20bd4836b36ffa5656a219333
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>

show more ...